Human-Centred Change Management for AI Adoption in Government: Adapting ADKAR for the Algorithmic State

Governments are introducing artificial intelligence (AI) into grievance handling, welfare targeting, health, revenue administration and many other functions. Experience with earlier e-government programmes shows that the human side of such change — whether officers understand, trust and use new tools appropriately — decides success as much as the technology. AI adds new challenges to familiar change-management problems: outputs are probabilistic rather than deterministic, the reasoning can be opaque, roles and discretion shift, and both over-reliance and under-use are risks. This paper adapts the widely used ADKAR model (Awareness, Desire, Knowledge, Ability, Reinforcement) to AI adoption in public administration. For each stage it identifies AI-specific barriers, practical interventions and indicators. It adds a sixth element — Accountability — to address the distinctive governance requirement that a human officer remains answerable for decisions supported by AI. The resulting "ADKAR+A" model is offered as a practical planning tool for programme leaders, illustrated with examples from Indian state e-governance contexts, and complemented by a ten-point readiness checklist. The paper draws on the author's experience in large public-sector digital transformation programmes and on published research; it is a conceptual contribution and does not report an evaluation.